Trinity Business School

Contact Us Trinity Business School
Trinity College
182 Pearse St
Dublin 2
D02 F6N2
Ireland

View the contact page for more contact and location information

business@tcd.ie

Ulrich Leicht-Deobald

“By examining fairness from multiple angles, the team has set new standards for socially responsible AI. True fairness means not only designing AI that serves society justly; it also means ensuring the systems we create are used in ways that reflect the values we wish them to uphold.”

Ambros Scope, PhD, Head of Leadership & Future of Work, Zurich Insurance Switzerland

Project Name: Socially Acceptable AI and Fairness Trade-offs in Predictive Analytics

Problem addressed by the research

The rapid adoption of AI has led to the critical question of how to design fair AI systems and applications. The question raises philosophical, technical, and psychological dilemmas: What does fairness mean? How is fairness perceived? How can fairness be implemented in AI?

Project description

To address these issues, Professor Ulrich Leicht-Deobald has been working with an interdisciplinary team of researchers from computer science, philosophy, economics, and organisational psychology. This interdisciplinary project, funded by the Swiss National Science Foundation, has developed a methodology for designing fair AI applications. The project connects the ethical, psychological and technological aspects of AI implementation. The team conducted experiments to examine lay people’s decisions on predictive accuracy versus group fairness trade-offs in AI-based personnel selection scenarios.

Project outcomes so far

The project developed an open-source FAIRNESS LAB audit tool to help stakeholders make informed decisions about tradeoffs between predictive accuracy and group fairness (https://joebaumann.org/FairnessLab/#/). Based on the project, a new teaching module for algorithmic fairness has been developed at the Zurich University of Applied Sciences, targeting computer scientists and data scientists. The material is being used in a dedicated undergraduate module on “Responsible AI” and in continuous education. Results fed into a recommendation report for the Swiss parliament and dialogue events with the public. The scientific results have been published in scholarly journals and presented at international conferences.

For more information